MAPK8IP3-Related Neurodevelopmental Disorder (NEDBA)

Neurodevelopmental disorder with or without variable brain abnormalities; MAPK8IP3-related neurodevelopmental disorder

Literature and regulatory information updated: 28 October 2026.

0. Initial Reference Table

1. Introduction

1.1. From axonal transport biology to human disease

The protein encoded by MAPK8IP3 was known for two decades before being associated with human disease. In 1999 and 2000, three independent lines of research described the same protein family: Ito and colleagues identified JSAP1 as a protein bound to the kinase JNK3; Kelkar and colleagues characterised Jip3 as a scaffold protein of the c-Jun N-terminal kinase (JNK) pathway, expressed mainly in the brain and accumulated in growth cones; and Bowman and colleagues described in Drosophila the sunday driver (syd) gene, whose product interacts with kinesin I through the kinesin light chain [9]. The name sunday driver sums up the fly phenotype: axonal cargo transport becomes slow and disorganised. In 2003, the Jip3-null mouse showed that the protein is indispensable for extrauterine life and for formation of the telencephalic commissure that connects the two hemispheres, a finding that anticipated the thin corpus callosum later observed in patients [23].

Before 2019, three de novo MAPK8IP3 variants had emerged in isolation in large sequencing studies of neurodevelopmental disorders: p.(Tyr94Cys) in a boy with autistic traits and an intelligence quotient (IQ) above 90 in an autism cohort, p.(Glu461Gly) in a 10-year-old girl with moderate ID, hypotonia and a previous clinical suspicion of Smith-Magenis syndrome, and p.(Arg1146Cys) in the Deciphering Developmental Disorders (DDD) study [17]. These findings were not sufficient to establish a gene–disease association.

1.2. The founding articles of 2019

The association was established in 2019 by two groups working in parallel. Platzer and colleagues, from the University of Leipzig, gathered through genomic data sharing (GeneMatcher) and international collaboration 13 unrelated individuals with heterozygous de novo variants, referred by centres in Brazil, Germany, Israel, Italy, the United States and the Netherlands, from a total cohort of 27,232 individuals with neurodevelopmental disorders studied by exome sequencing [17]. The number of de novo variants observed exceeded that expected from the mutation rate of the gene (binomial test, p = 6.165×10⁻⁵), and structural modelling and assays in the nematode Caenorhabditis elegans provided the first functional evidence [17]. The title of the article established the central idea of the entity: ID with variable brain abnormalities.

Iwasawa and colleagues published in the same year, in the Annals of Neurology, five Japanese patients from four families with two recurrent variants, p.(Arg578Cys) and p.(Arg1146Cys), a more severe phenotype (spastic diplegia, severe to profound ID, cerebral atrophy and hypoplasia of the corpus callosum) and the demonstration of axonal varicosities in zebrafish embryos overexpressing the mutant human protein [1, 24]. The OMIM entry #618443 was created on 22 May 2019 on the basis of these two series [1].

1.3. Subsequent milestones

The following years broadened the clinical and mechanistic spectrum. In 2022, ClinGen classified the relationship between MAPK8IP3 and NEDBA as definitive [11]. In 2023, Kárteszi and colleagues proposed an autosomal recessive lower motor neuron disease phenotype in an infant with severe congenital hypotonia resembling lethal spinal muscular atrophy (SMA) [14]. In 2025, Zhang and colleagues showed, in fibroblasts and neurons derived from induced pluripotent stem cells (iPSCs) from a patient with p.(Arg578Cys), a toxic gain-of-function mechanism and reversed it with ASOs [25]; in the same year a study of personalised ASO treatment in a single participant began [19]. The largest clinical series, with 32 individuals assessed in a natural history study in the United States, was published online in July 2025 and in print in 2026 [3]. In 2026 the first family with autosomal dominant transmission over three generations was described [13], and a preprint described the first knock-in mice with human variants in different domains of the protein [26].

1.4. Nomenclature and definition of the entity

OMIM names the disorder neurodevelopmental disorder with or without variable brain abnormalities (NEDBA) [1], a designation also adopted by MedGen and, through the identifier MONDO:0032755, by ClinGen [2, 11]. The recent clinical literature and patient associations prefer «MAPK8IP3-related neurodevelopmental disorder» [3, 4]. The acronym NEDMAGA, sometimes mistakenly associated with this gene, belongs to the neurodevelopmental disorder with movement abnormalities, abnormal gait and autistic features caused by the recurrent ZSWIM6 variant p.(Arg913Ter) (OMIM #617865) [6, 7]; the two entities share clinical features, but have different genes, brain imaging and facial phenotype (section 6). Orphanet has not assigned a dedicated code to NEDBA and groups it in the heterogeneous category of non-specific syndromic ID (ORPHA:528084), defined by the association of ID with other variable anomalies without a well-characterised syndrome [15].

NEDBA is defined as a neurodevelopmental disorder with onset in infancy or early childhood, caused by heterozygous pathogenic MAPK8IP3 variants, with global developmental delay, late-acquired walking, ID of variable degree and poor or absent speech; hypotonia, spasticity and ataxia are frequent, about half of patients have abnormalities on brain MRI, epileptic seizures are not prominent and there is no common facial phenotype [1]. The entity has no numbered subtypes and does not belong to a syndromic class with types; it belongs to the broader group of disorders of axonal transport and of molecular motor adaptors, which includes the disorders of KIF1A, KIF5A, KIF5C, DYNC1H1 and BICD2 [17, 27].

2. Molecular Genetics and Pathophysiology

2.1. The gene and the JIP3 protein

MAPK8IP3 is located at 16p13.3 and encodes JIP3 (JNK-interacting protein 3), also termed JSAP1 or SYD2 [9, 13]. The MANE Select transcript is NM_001318852.2 (ENST00000610761.2), with the coding sequence between nucleotides 146 and 4159 [12]. The protein is organised into four functional regions described in the founding article: an RILP homology domain 1 (RH1) in the N-terminal region, a homodimeric leucine zipper (LZ), a second RILP homology domain (RH2) and a WD40 domain in the C-terminal region [9, 17]. The gene is highly intolerant to variation: probability of loss-of-function intolerance (pLI) of 1.00, LOEUF (loss-of-function observed/expected upper bound fraction) of 0.41 in gnomAD v4.0, missense z-score of 4.06 in ExAC and probability of haploinsufficiency (pHaplo) of 0.93 [11, 13, 17]. JIP3 expression in the mouse brain increases between embryonic day 10 and postnatal day 10, the period of migration and axonal growth [17].

2.2. Cellular functions

JIP3 has two families of functions. The first is that of a scaffold protein of the JNK pathway: it binds JNK and the components of the mitogen-activated protein kinase cascade and assembles them into a signalling module [10]. The second is that of an adaptor for the molecular motors of axonal transport. Axonal transport is bidirectional and ATP-dependent: anterograde transport, from the cell body to the terminal, is driven by kinesins and delivers RNA, proteins and organelles to synapses; retrograde transport depends on dynein and is essential for autophagic-lysosomal degradation and for neurotrophic factor signalling [27]. JIP3 binds the kinesin-1 heavy and light chains, promotes motility along microtubules, bridges the brain-derived neurotrophic factor (BDNF) receptor, NTRK2/TrkB, and the kinesin light chain KLC1, locally activates cofilin at axon tips and is required for axonal elongation and regeneration and for radial migration of cortical neurons [10, 17].

Studies in C. elegans, mice and human neurons have shown a second function, independent of kinesin-1 and conserved through evolution: the «clearance» of organelles that do not belong in the axon. Without the orthologue unc-16, lysosomes, early endosomes and Golgi components enter the axon aberrantly; in the JIP3-null mouse, lysosomes accumulate in the axon [17]. In Jip3-deficient neurons, lysosomes accumulate in axonal swellings containing BACE1 and presenilin 2, with increased production of β-amyloid peptide, and JIP3 haploinsufficiency worsens plaque deposition in a mouse model of Alzheimer's disease [28]. In iPSC-derived human neurons, JIP3 and its paralogue JIP4 (SPAG9) have overlapping roles in the axonal transport of lysosomes [29]. An ultrastructural study from 2026 added a somatic function: iPSC neurons lacking JIP3 accumulate immature lysosomes and lose mature lysosomes in the cell body, with delayed trafficking of endocytic cargo, indicating that JIP3 regulates endolysosomal maturation outside the axon [30].

The coordination between the two motors was detailed by a group at i3S, in Porto: the dynein light intermediate chain binds the RH1 domain of JIP3 and JIP4; a separation-of-function mutation in RH1 that abolishes this binding, without affecting kinesin binding, causes in C. elegans accumulation of endolysosomal organelles at the neurite tip, and this defect is corrected by a disease-associated mutation in the leucine zipper that eliminates binding to the kinesin light chain [31]. JIP3 thus coordinates bidirectional transport by dynein and kinesin-1. In 2026 it was shown that JIP3 and JIP4 bind, through RH1, the CC4 domain of the kinesin KIF5B on a surface distinct from the dynein-binding surface, with affinities that differ about 40-fold between the two proteins; this mechanism contributes to relieving kinesin-1 autoinhibition [32].

2.3. Animal models

The Jip3⁻/⁻ mouse cannot breathe and dies shortly after birth, with severe defects of telencephalic morphogenesis, absence of the telencephalic commissure and reduced RhoA/ROCK signalling [23]. The heterozygous Mapk8ip3⁺/⁻ mouse is viable and fertile without evident abnormalities, the double knockout of Mapk8ip3 and its paralogue Jlp in the dorsal telencephalon causes axonal dystrophy and progressive neuronal degeneration, and the heterozygous rat has shorter hippocampal axons [13].

In Platzer's assays in C. elegans, the variant equivalent to p.(Tyr37Ter) increased axonal lysosome density about 88-fold relative to the wild type (slightly more than the reference null mutant) and the equivalent of p.(Leu444Pro) about 15-fold; the other four missense variants tested did not significantly alter this parameter [17]. In locomotion (swimming cycles per minute, 72 animals per strain), the null mutant swam at about 35% of the wild-type rate, p.(Tyr37Ter) behaved like the null, p.(Leu444Pro) reduced the rate to about 50%, p.(Tyr94Cys), p.(Arg525Gln) and p.(Glu461Gly) to 93–94% of their respective revertants, and p.(Arg1146Cys) had no significant effect; CRISPR reversion to the wild type normalised the phenotypes [17]. The authors noted a paradox: the variants with the greatest effect in the assays were not those with the worst cognitive outcome, and the most recurrent variant, p.(Arg1146Cys), had no effect in the organelle clearance assay, which suggested disruption of kinesin-1-dependent functions that the model did not test [17]. In zebrafish, overexpression of the human variants p.(Arg578Cys) and p.(Arg1146Cys) produced axonal varicosities in the posterior lateral line nerve [24].

A preprint from July 2026, not yet peer reviewed, described knock-in mice with variants in three domains. Homozygotes for p.E27X and p.R1147C (the murine equivalent of p.Arg1146Cys) are embryonic lethal; heterozygotes are viable, with variant-dependent phenotypes: p.R579C (equivalent of p.Arg578Cys) reduces locomotion and causes clasping suggestive of hypertonia, p.R1147C causes hyperactivity, clasping and lower brain weight, and p.E27X worsens rotarod performance without altering gait; the two missense variants reduce anxiety-like behaviour, and no line showed degeneration up to 12 months [26].

2.4. Mechanisms: haploinsufficiency and toxic gain of function

Current data point to two mechanisms coexisting in the same gene. Truncating variants in exon 1 and canonical splicing variants are predicted to lead to nonsense-mediated mRNA decay and act through haploinsufficiency [13, 17]. This prediction has not been demonstrated experimentally: in the family with the variant c.2630-2_2630-1del, RNA analysis did not show degradation of the aberrant transcript, and neither allelic expression studies nor studies with decay inhibition were performed, and the authors acknowledge that haploinsufficiency may not be the only mechanism, which is consistent with the existence of asymptomatic carriers of loss-of-function variants [13, 14]. Some recurrent missense variants, in contrast, have a toxic gain-of-function effect. Zhang and colleagues showed in fibroblasts and iPSC-derived neurons from a patient with p.(Arg578Cys) that the mutant protein alters the interactome and subcellular localisation of JIP3, disrupts the motility of late endosomes along microtubules without the lysosomal defect typical of loss-of-function models, increases JNK activation with apoptosis and reduced proliferation and viability, and uncouples dopamine D1 receptor signalling, without affecting D2, through increased interaction with that receptor [25]. The patient's neurons also had increased NLRP2 and reduced JIP4, which reduces compensation by the paralogue [25].

The distinction has clinical and therapeutic consequences. In the largest clinical series, missense variants were associated with more severe symptoms than predicted loss-of-function variants [3], and the murine preprint proposed that missense variants act mainly through toxic gain of function and haploinsufficiency variants cause milder phenotypes; the logical therapy would be reducing expression in the first case and increasing it in the second [26]. Not all missense variants have been studied, and the mechanism of a novel variant should not be presumed from its class alone.

2.5. Variant spectrum and nomenclature

The founding series used transcript NM_015133.4, whose numbering coincides with MANE up to codon 94 and differs from codon 400 onwards by +3 nucleotides (+1 amino acid) [12]. The two recurrent variants most often cited as p.Arg578Cys and p.Arg1146Cys correspond, in MANE, to c.1735C>T p.(Arg579Cys) and c.3439C>T p.(Arg1147Cys). The following conversions were confirmed in VariantValidator and the classification status in ClinVar on 28 October 2026 [5, 12].

On the date of consultation, ClinVar held 752 records for MAPK8IP3, 69 of them classified as pathogenic or likely pathogenic, including copy number variations (CNVs) [5]. The additional sequence variants, in MANE nomenclature, include truncating variants distributed along the gene (e.g., c.1363G>T p.(Glu455Ter), c.1402C>T p.(Gln468Ter), c.1708A>T p.(Lys570Ter), c.2214_2215del p.(Leu739fs), c.2919G>A p.(Trp973Ter), c.3334_3335del p.(Leu1112fs), c.3559C>T p.(Arg1187Ter)), canonical splicing variants (c.318+1G>A, c.748-2A>C, c.2121+1dup, c.2122-1G>A, c.3238-1G>A), missense variants (c.167T>A p.(Met56Lys), c.2401T>G p.(Phe801Val), c.3257C>T p.(Pro1086Leu)) and a variant in a microsatellite (NC_000016.10:g.1767914AG[2]); most have a single submitter [5].

Two discrepancies in the recent literature should not be reproduced. Zhang 2025 describes the recurrent RH2 domain variant as «c.1714C>T» and cites p.(Glu461Gly) as «c.1364A>G p.(Glu455Gly)», a notation repeated by Yi 2026; neither of these forms corresponds to NM_015133.4 or to MANE, and the correct designations are, respectively, c.1735C>T p.(Arg579Cys) and c.1385A>G p.(Glu462Gly) in MANE [12, 13, 25]. GeneGuidance and some laboratory reports use a third transcript, NM_001040439, which makes it necessary always to confirm the transcript before comparing variants [10, 34]. In the mouse models, R579C and R1147C designate the residues equivalent to the recurrent human variants, whose numbering in MANE is the same [26].

2.6. Genotype–phenotype correlations

The available correlations rest on small numbers. In the founding series, the two people with p.(Leu444Pro) had bilateral perisylvian polymicrogyria with an otherwise normal brain; the two with p.(Arg578Cys) had cerebral atrophy, loss of white matter volume, a very thin corpus callosum, cerebellar atrophy or hypoplasia, spasticity, microcephaly and short stature; two of three with p.(Arg1146Cys) had white matter loss and a thin corpus callosum, spasticity and cortical visual impairment; and two of three with truncating variants had an ataxic gait [17]. In the Japanese series, carriers of the two recurrent missense variants had severe to profound ID and the three oldest never walked [1]. In a literature review published in 2026, carriers of p.(Arg578Cys) and p.(Arg1146Cys) had moderate ID in 30%, severe in 60% and profound in 10%, and epileptic seizures had been described only in six missense carriers, four of them with p.(Arg578Cys) [13]. In the series of 32 individuals, missense variants (62.5%) were associated with greater severity than predicted loss-of-function variants (34.4%) [3]. The variant profiles of the natural history cohort are described in section 3.5 [26].

2.7. Inheritance, penetrance and expressivity

The classic form is autosomal dominant, and the 13 variants of the founding series were de novo, confirmed by trio sequencing in 12 individuals [17]. Two Japanese siblings with p.(Arg578Cys), children of parents without the variant, suggest parental germline mosaicism [1]. In 2026, a Chinese family demonstrated vertical transmission over three generations of a splicing variant with highly variable expressivity: mild learning difficulties in the mother, language delay and learning difficulties in the sister, severe global delay with autism spectrum disorder (ASD) in the proband and a carrier grandmother whose development was unknown [13]. The authors mention carriers of loss-of-function variants without apparent disease, which suggests incomplete penetrance for some alleles [13].

There is also a proposal of autosomal recessive inheritance. Kárteszi and colleagues described a patient with severe congenital hypotonia of neuronal origin, mimicking lethal SMA, compound heterozygous for a splicing variant and a missense variant inherited from healthy parents, with a reduction of more than 60% in JIP3 on muscle biopsy [14]. SysNDD records this association as a separate autosomal recessive entity («neuromuscular disease», MONDO:0019056) without a validity classification («not applicable»), in contrast with the Definitive category of the dominant form [16]. The two patterns coexist in the sources, but with very different strength of evidence: the dominant one rests on dozens of patients and a definitive curation, the recessive one on a single case.

2.8. Copy number variations and other associations

MAPK8IP3 lies in a gene-rich region of 16p13.3. ClinVar contains deletions that include MAPK8IP3 and TSC2, submitted as tuberous sclerosis, terminal 16p13.3 deletions and extensive duplications of the region (including 16p13.3 duplication syndrome), classified as pathogenic; in these CNVs the isolated contribution of MAPK8IP3 cannot be separated from that of the other genes [5]. Beyond NEDBA, MAPK8IP3 has emerged as a candidate gene in an Austrian cohort of early-onset dementia (60 patients studied by exome sequencing) and in a family with obsessive-compulsive disorder in which a missense variant co-segregated with a DLGAP1 deletion [35, 36]; these are preliminary associations, without demonstration of causality.

3. Clinical Spectrum and Manifestations by System

3.1. Overview

NEDBA presents in infancy or early childhood with global developmental delay, almost always with hypotonia, and evolves into ID of variable degree, poor or absent speech and abnormalities of tone and coordination [1]. The sources describe the disorder in four sets of patients that should not be added together, because they partially overlap and were assembled with different criteria: the founding series of 13 individuals [17], the Japanese series of five [1, 24], the natural history series of 32 [3] and a subset of 18 individuals from that natural history study with detailed motor assessment by genotype [26]. All available data describe patients without disease-modifying treatment; natural history and treated disease therefore cannot be separated.

3.2. Founding series (Platzer 2019)

The 13 individuals were aged between 3 and 19 years at assessment [17]. All had psychomotor developmental delay; ID was mild in 4 (31%), moderate in 6 (46%) and severe in 3 (23%), and the three individuals with a measured IQ had values of 48 to 49. Eleven had abnormalities of tone (hypotonia in 9, spasticity in 4, both in 2), two had ataxia and one an unsteady gait. Two had ASD, two cortical visual impairment, three scoliosis, two short stature and three microcephaly (−2.5, −2.5 and −2.7 standard deviations). One individual had recurrent generalised seizures, but also had a de novo variant in SLC6A1, a gene of myoclonic-atonic epilepsy, which makes a dual monogenic diagnosis likely; three others had a single generalised seizure, two of them in a febrile context. None had regression and there was no consistent facial dysmorphism [17].

SD, standard deviations. Source: Table 1 and Figure 1 of Platzer et al. 2019 [17]; in MANE, p.(Arg578Cys) = p.(Arg579Cys) and p.(Arg1146Cys) = p.(Arg1147Cys).

3.3. Japanese series (Iwasawa 2019)

The five Japanese patients, from four families and aged 5 to 29 years, had a more severe picture [1]. All had hypotonia in infancy; only one 5-year-old girl acquired walking (at 2 years), another 5-year-old child moved with lateral support, and the three oldest had spastic diplegia, never walked and were wheelchair-dependent. Most did not speak or said few words, ID was severe or profound in all (developmental quotients of 27 to 34) and the youngest had autistic traits. Two had epilepsy and three had spikes on the electroencephalogram (EEG). MRI was abnormal in all, with cerebral atrophy and hypoplasia of the corpus callosum, and three had delayed myelination. Most had non-specific dysmorphic features (round face, prominent nasal root, thin upper lip). Two had short stature, three obesity and the two siblings (a boy and a girl) precocious puberty. In one patient, sural nerve biopsy showed loss of myelinated fibres, electromyography a neurogenic pattern and conduction studies an axonal neuropathy [1]. The severity of this series may reflect the selection of carriers of the two recurrent missense variants [24].

3.4. Natural history series (Sudnawa 2026)

The largest series, from a natural history study conducted at Boston Children's Hospital and Columbia University, included 32 individuals with heterozygous pathogenic or likely pathogenic variants, with a median age of 7.5 years (1.3 to 22.0) [3]. The variants were missense in 62.5% and predicted loss of function in 34.4%. The most common symptoms were cognitive impairment, hypotonia, motor difficulties, strabismus, microcephaly and attention deficit; thinning of the corpus callosum was described in 62.1% (the abstract does not give the denominator). Almost all walked independently, but with lower gait speed and a wider base of support than controls. General conceptual ability on the Differential Ability Scales, second edition (DAS-II), had a mean of 62.5 ± 26.5, reflecting a range from moderate ID to near-average intellectual functioning. Quantitative EEG analysis showed only a trend towards a lower peak power frequency compared with typically developing children [3]. The article also assessed longitudinal adaptive functioning; the values and the frequencies of each sign are not given in the abstract, the only text accessible (see limitations in section 9.4).

The contrast between the Japanese series, in which three of five never walked, and the natural history series, in which almost all walked, shows the effect of recruitment: the 2026 series included one-third loss-of-function variants, associated with milder pictures, and was recruited with the support of patient associations, whereas the 2019 series gathered patients identified in a diagnostic setting [1, 3, 4].

3.5. Profiles by genotype (Crowder 2026, preprint)

In the same natural history study, 18 individuals were assessed in person with motor scales (Gross Motor Function Classification System, GMFCS; Gross Motor Function Measure, GMFM-88/66; Box and Block; nine-hole peg; 10-metre and 6-minute walk tests; instrumented walkway) and cognitive scales (Bayley-4, DAS-II or WASI-II) [26]. All had gross motor and language delay. The results by genotype, from a preprint not peer reviewed, are summarised below.

Source: Crowder et al. 2026, Table 2 [26]. Percentages over denominators of 5 and 12; differences between groups without formal statistical testing in the text consulted.

The pattern suggests two profiles: p.(Arg578Cys) with microcephaly, hypertonia and abnormal movements (tremor, myoclonus, dystonia), and p.(Arg1146Cys) with more frequent hypotonia, ADHD and ASD. The profiles coincide with the phenotypes of the corresponding knock-in mice (section 2.3) [26], but the groups are small.

3.6. Manifestations by system

Neurodevelopment and cognition. Global delay is constant. ID ranges from mild to profound; in the founding series moderate ID predominated [17], in the Japanese series severe to profound ID [1] and in the series of 32 the mean general conceptual ability was 62.5 [3]. Language is the most affected domain: many patients do not acquire useful speech or remain at single words [1, 13]. One carrier of p.(Tyr94Cys) had an IQ above 90 and autistic traits, which shows that the spectrum includes intellectual functioning within the average range [17]. Regression is not characteristic: none of the 13 patients in the founding series regressed [17], but a 30-month-old girl with the intronic variant c.2820-3C>G had developmental regression, so the diagnosis should not be excluded in the presence of regression [33].

Motor function. Independent walking is acquired late by most, with slow, broad-based gait [3]; some carriers of recurrent missense variants have spastic diplegia and do not walk [1]. Hypotonia predominates in loss-of-function variants and in p.(Arg1146Cys), hypertonia in p.(Arg578Cys) [26]. Ataxia has been described mainly with truncating variants [17], and tremor, myoclonus and dystonia with p.(Arg578Cys) [25, 26]. In an Australian population-based cerebral palsy cohort, a child with a white matter injury pattern on MRI (periventricular leucomalacia) had a MAPK8IP3 variant, which shows that the disorder can present as spastic cerebral palsy with apparently lesional imaging [37].

Neuroimaging. About half to two-thirds of patients have MRI abnormalities [1, 3]. In the founding series, 8 of 12 MRI scans were abnormal: a very thin or short corpus callosum in 5 (3 with cerebral atrophy and 4 with considerable white matter loss), isolated bilateral perisylvian polymicrogyria in 2 and mild cerebellar atrophy with hyperintensity of the internal capsules in 1 [17]. In the Japanese series all had cerebral atrophy and hypoplasia of the corpus callosum, and three had delayed myelination [1]. In the series of 32, a thin corpus callosum was the most frequent finding (62.1%) [3]. The perisylvian polymicrogyria linked to p.(Leu444Pro) is probably a post-migrational disorder [17]. A normal MRI is compatible with the diagnosis; the family with dominant transmission had normal MRI [13].

Epilepsy and EEG. Seizures are not prominent [1]. In the founding series there were three isolated seizures, two febrile, and one patient with epilepsy who had a second causal variant [17]; in the Japanese series two of five had epilepsy and three had spikes on the EEG [1]. The seizures described cluster in missense carriers, especially p.(Arg578Cys) [13]. NORD mentions slow waves with spikes on the EEG [4]. Quantitative EEG showed no diagnostic abnormalities [3].

Behaviour. Attention deficit, ASD, anxiety and sensory integration abnormalities are frequent, predominantly in carriers of p.(Arg1146Cys) and of truncating variants [3, 26]. ASD may be the reason for referral: in the family with dominant transmission, the proband was diagnosed with autism at 2 years, before the genetic diagnosis [13].

Vision. Strabismus is one of the most common signs in the series of 32 [3], and cortical visual impairment was described in 2 of 13 in the founding series [17]. One of the cases with regression had esotropia [33].

Growth and endocrinology. Microcephaly occurs in 3 of 13 in the founding series and in 80% of carriers of p.(Arg578Cys) in the genotype cohort [17, 26]. Short stature, obesity and precocious puberty were described in the Japanese series [1], and NORD includes precocious puberty and short stature among the additional signs [4]. Relative macrocephaly (+1.7 SD) was recorded in the proband of the Chinese family [13].

Musculoskeletal system. Scoliosis occurred in 3 of 13 individuals in the founding series [17].

Peripheral nervous system. One Japanese patient had axonal neuropathy proven by biopsy and electrophysiology [1], and the proposed recessive form is a lower motor neuron disease [14]. The frequency of subclinical neuropathy has not been studied systematically.

Face and other organs. There is no common facial phenotype; the features described (round face, prominent nasal root, thin upper lip) are non-specific [1, 4]. NORD mentions reflux and feeding difficulties that justify feeding therapy [4]. The Chinese proband had an inguinal hernia, without malformations [13].

3.7. Proposed recessive form

The only case of recessive inheritance described had severe congenital muscular hypotonia of neuronal origin, mimicking lethal SMA type 1; the splicing variant and the missense variant were inherited from healthy parents, and muscle biopsy showed a reduction of more than 60% in JIP3 [14]. The age at presentation, the exact variants and the course are not given in the abstract, the only text accessible.

3.8. Phenotypes outside the neurodevelopmental disorder

The inclusion of MAPK8IP3 among the candidate genes for early-onset dementia in an Austrian cohort and the co-segregation of a missense variant with obsessive-compulsive disorder in a multiplex family are preliminary signals, without replication [35, 36]. The role of JIP3 in lysosome transport and in β-amyloid production in Alzheimer's disease models gives biological plausibility to the first hypothesis, but does not demonstrate increased risk in human carriers [28].

4. Clinical Recognition and Warning Signs

4.1. When to suspect the disorder

NEDBA has no pathognomonic sign and no recognisable facial phenotype, and the diagnosis almost always emerges from a genomic test requested for developmental delay or ID [1, 17]. Some contexts make the hypothesis more likely and justify including MAPK8IP3 in the analysis or reanalysing an older exome.

4.2. Warning signs

No disease-specific emergency is described in the sources. The signs that warrant prompt assessment are those of the comorbidities: epileptic seizures, especially in missense carriers [13]; loss of acquired skills, which is atypical and requires investigation of other causes, although it has been described in one case [17, 33]; respiratory difficulty and severe hypotonia in the infant, in the context of the recessive form [14]; and dysphagia or respiratory difficulty after botulinum toxin injection, which may reflect spread of the toxin (section 7.3) [38].

4.3. Newborn screening

There is no biochemical newborn screening and no newborn genomic screening programme directed at this disorder in the sources consulted.

5. Diagnosis: Criteria, Tests and Algorithms

5.1. Criteria

There are no formal clinical criteria and no biochemical or functional biomarker [10]. The diagnosis is molecular: it is established in an individual with a compatible neurodevelopmental disorder and a heterozygous pathogenic or likely pathogenic MAPK8IP3 variant, classified according to the criteria of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology (ACMG/AMP) [1, 39]. No methylation signature (episignature) was identified in the sources consulted.

5.2. Initial clinical assessment

The assessment includes developmental history, neurological examination with attention to tone, gait and abnormal movements, head circumference, ophthalmological assessment (strabismus, visual function) and brain MRI [3, 10]. MRI is the main supporting investigation, but a normal MRI does not rule out the diagnosis [10, 13]. EEG is indicated in the presence of paroxysmal events; quantitative EEG has no diagnostic value [3]. Nerve conduction studies and electromyography are indicated in the presence of weakness, areflexia or severe hypotonia, given the axonal neuropathy described in one patient and the recessive lower motor neuron form [1, 14].

5.3. Genetic testing

The 2021 ACMG evidence-based guideline strongly recommends that exome or genome sequencing be considered as a first- or second-line test in children with congenital anomalies with onset before 1 year of age or with developmental delay or ID with onset before 18 years, because it has a higher diagnostic yield than classical tests and is possibly more cost-effective when ordered early [18]. For NEDBA, which has no recognisable phenotype, this is the usual route to diagnosis: 12 of the 13 patients in the founding series were diagnosed by trio exome sequencing [17]. An ID panel that includes MAPK8IP3 is an alternative when exome sequencing is not available, and microarray identifies 16p13.3 deletions that encompass the gene, although in that case the phenotype depends on the other genes included [5].

5.4. Limitations of testing

A negative result does not exclude the disorder. Exome sequencing may fail to detect intronic variants outside the canonical junctions: the variant c.2820-3C>G was identified by genome sequencing in a child with a non-diagnostic exome from 2019 [33]. Exomes analysed before the 2019 publication did not report MAPK8IP3, and reanalysis would have sufficed in that case [33]. The exome has limited sensitivity for small intragenic CNVs, and low-level mosaicism in the parents may not be detected by sequencing of blood, as suggested by the Japanese sibling pair [1]. The proportion of missense (62.5%) and loss-of-function variants (34.4%) in the series of 32 describes the observed spectrum, not the sensitivity of any method [3].

5.5. Variant interpretation

Most pathogenic variants are private; the exceptions are p.(Arg578Cys) and p.(Arg1146Cys), recurrent in several series and classified as pathogenic by multiple submitters without conflicts [5, 9]. Confirmation of de novo occurrence by parental testing is the most useful criterion, given that none of the variants in the founding series was in gnomAD and all missense variants affected residues conserved down to zebrafish [17].

Segregation requires care in both directions. In a duo exome, the variant c.2366C>T p.(Thr789Met) (NM_001040439.2), inherited from the unaffected father, was reclassified as likely benign, and the real diagnosis was a de novo ATP1A3 variant [34]. In the Chinese family, conversely, the pathogenic variant was inherited from a mother with mild learning difficulties [13]. Inheritance from an apparently healthy parent weighs against pathogenicity only after the phenotype of that parent has been assessed.

Splicing variants benefit from RNA studies. In the Chinese family, leucocyte RNA sequencing showed aberrant splicing in 44% (23/52), 58% (29/50) and 46% (18/39) of the reads of the mother, the proband and the sister; RT-PCR, cDNA sequencing and cloning confirmed activation of a cryptic acceptor in exon 22, with a 44-nucleotide deletion (r.2630_2673del) and a frameshift, which allowed the variant to be classified as likely pathogenic (PVS1, PM2_supporting) [13]. The authors did not study mRNA decay and recommend functional assays to interpret novel variants [13]. For c.2820-3C>G, prediction tools indicated loss of the acceptor (MaxEntScan from 9.35 to −8.98; SpliceAI 0.70) [33].

5.6. Report content

The report should state the gene, the transcript with version (MANE NM_001318852.2), the variant according to Human Genome Variation Society (HGVS) nomenclature at cDNA and protein level, the zygosity, the ACMG/AMP classification with the criteria applied, the origin (de novo, inherited or unknown) and, when the variant has been published under another numbering, the correspondence with the historical NM_015133.4 nomenclature [12, 39]. For p.(Arg579Cys) and p.(Arg1147Cys), explicitly stating the historical equivalent (p.Arg578Cys, p.Arg1146Cys) avoids the false impression of a novel variant.

5.7. Practical algorithm

5.8. Prenatal and preimplantation diagnosis

The phenotype has no known prenatal expression that would allow ultrasound suspicion; the sources describe no fetal findings. When the familial variant is known, prenatal diagnosis by chorionic villus sampling or amniocentesis and preimplantation genetic testing are technically possible, under the general conditions applicable to monogenic diseases (section 8.3).

6. Differential Diagnosis

6.1. Comparative table

6.2. NEDMAGA and ZSWIM6

NEDMAGA is the closest differential by name and by several clinical features. It was described in 2017 in seven unrelated individuals with the same de novo nonsense variant c.2737C>T p.(Arg913Ter) (NM_020928.2) in the penultimate exon of ZSWIM6, which escapes mRNA decay and produces a truncated protein lacking the Sin3-like domain, with a proposed dominant-negative effect [7]. The missense variant p.(Arg1163Trp) in the same gene causes, in contrast, acromelic frontonasal dysostosis [7]. NEDMAGA shares with NEDBA severe ID, hypotonia, late and broad-based gait, absence of useful speech, autistic traits and, in some patients, spasticity and peripheral neuropathy (14%) [8]. It is distinguished by normal MRI, recurrent reflux vomiting (86%), hand stereotypies and tics, happy demeanour and a recognisable facial phenotype [8]. A Brazilian case from São Paulo illustrates the picture: an 8-year-old boy with neonatal hypotonia, walking at 3 years, absence of useful speech, fundoplication for intractable vomiting at 2 years, stereotypies and tics from 5 years and a tonic-clonic episode at 7 years, with normal MRI, EEG, metabolic screening, karyotype, microarray and fragile X testing, diagnosed by exome sequencing [8]. Figure 1 of that report shows the facial phenotype (https://pmc.ncbi.nlm.nih.gov/articles/PMC10883404/figure/mdc313954-fig-0001/).

6.3. Other axonal transport disorders

The genes of molecular motors and their adaptors cause a spectrum ranging from late-onset motor diseases (motor neuropathies, Charcot-Marie-Tooth disease, spastic paraplegias, amyotrophic lateral sclerosis) to early-onset encephalopathies with abnormalities of cortical migration [27]. Within this group, NEDBA comes closest to KIF1A-associated neurological disorder, which NORD lists as a differential [4]. The clinical distinction rests on the degenerative component of KAND (optic atrophy, progressive spastic paraparesis, neuropathy and progressive cerebellar atrophy), absent in NEDBA, in which regression is not described and the mice do not degenerate up to 12 months [17, 26, 40]. DYNC1H1 disorders combine ID and cortical malformations with proximal axonal motor neuropathy, which calls for conduction studies when there is hypotonia or weakness [42]. Bilateral perisylvian polymicrogyria links NEDBA to DYNC1H1, BICD2, KIF5C and acquired causes, and imaging alone does not distinguish these entities [17, 41].

6.4. Cerebral palsy and congenital hypotonia

In an Australian population-based cohort of 331 children with cerebral palsy born between 1986 and 2018, a genetic diagnosis was achieved in 24% (80/331); 19% of the children had white matter injury, and one of them, with periventricular leucomalacia, had a MAPK8IP3 variant [37]. The authors conclude that the MRI pattern should not preclude genomic testing in cerebral palsy [37]. In the infant with severe congenital hypotonia and suspected SMA, normal SMN1 dosage should lead to exome sequencing, which may identify the proposed recessive form [14].

7. Treatments and Risk Management Strategies

7.1. General principles and guidelines

There is no NEDBA-specific management guideline or consensus; searches of Europe PMC, PubMed, NORD, ERN ITHACA and ClinicalTrials.gov identified no document of this type, and there is no GeneReviews chapter [4, 10, 20]. The National Organization for Rare Disorders (NORD) report, written with investigators from the natural history study, describes symptom-directed and supportive treatment, focused on the development of speech, motor function and social skills, with occupational therapy, physiotherapy, speech therapy and feeding therapy directed at specific delays, and recommends genetic counselling for patients and families [4]. The medicines described below treat comorbidities (ADHD, spasticity) and are used within the general indications for these problems; none has been studied in NEDBA, and the doses come from the official labels, as a declared extrapolation from paediatric ADHD and from paediatric spasticity of cerebral origin (cerebral palsy).

7.2. Non-pharmacological interventions

Rehabilitation is the foundation of treatment. Physiotherapy addresses hypotonia, spasticity, ataxia and the slow, broad-based gait, which persists even in patients who walk [3, 4]. Occupational therapy treats fine motor delay, present in more than 80% of carriers of the recurrent missense variants [26]. Speech therapy and augmentative and alternative communication address absent or minimal speech, the most affected domain [1, 4]. Feeding therapy is indicated for feeding difficulties [4]. Behavioural intervention and educational support address ASD, ADHD, anxiety and sensory integration abnormalities [13, 26]. In the proband of the Chinese family, rehabilitation started after the diagnosis of autism had limited benefit before the genetic diagnosis [13]. The scales used in the natural history study and in the ASO trial (GMFCS, GMFM-88, Manual Ability Classification System (MACS), Bayley-4, Vineland-3) also serve to measure progress in clinical practice [19, 26].

Correction of strabismus, assessment of cortical visual impairment, monitoring and orthopaedic treatment of scoliosis and endocrinological treatment of precocious puberty follow the general indications for these problems, because there are no NEDBA-specific data [1, 3, 17].

7.3. Symptomatic pharmacological treatment

Methylphenidate for ADHD. Attention deficit is common in the series of 32 and ADHD affects half of p.(Arg1146Cys) carriers in the genotype cohort [3, 26]. Methylphenidate is a central nervous system stimulant that inhibits dopamine and noradrenaline reuptake. In Portugal, Rubifen® (5, 10 and 20 mg tablets) is indicated for ADHD from 6 to 18 years and is not authorised in adults with ADHD; it is started at 5 mg once or twice daily (at breakfast and lunch), with weekly increments of 5 to 10 mg, up to a maximum of 60 mg/day, and is stopped if there is no improvement after one month of titration [44]. The FDA label for Ritalin® provides the same regimen from 6 years of age (5 mg twice daily, weekly increments of 5 to 10 mg, doses above 60 mg/day not recommended) and includes adults [45]. In Brazil, methylphenidate has active ANVISA registration, among others as Ritalina® and Concerta® [46]. Before starting, cardiac disease (personal history, family sudden death, arrhythmias), tics and psychiatric disease are assessed; methylphenidate is contraindicated with monoamine oxidase inhibitors and within 14 days of their discontinuation, and the Portuguese label also contraindicates it in glaucoma, phaeochromocytoma, hyperthyroidism and pre-existing cardiovascular or cerebrovascular disorders [44, 45]. Common adverse effects are decreased appetite, weight loss, insomnia, anxiety, headache, tachycardia and abdominal pain; the FDA label has a boxed warning on abuse, misuse and addiction [45]. Methylphenidate may lower the seizure threshold in patients with a history of seizures or EEG abnormalities, which matters in a disorder with occasional seizures and electroencephalographic abnormalities [44]. Monitoring is described in section 11.

Oral baclofen for generalised spasticity. Hypertonia affects 60% of p.(Arg578Cys) carriers and spastic diplegia was described in the Japanese series [1, 26]. Baclofen is a GABA-B receptor agonist that reduces reflex transmission in the spinal cord. In Portugal, Lioresal® (10 and 25 mg tablets) has a paediatric indication for symptomatic treatment of spasticity of cerebral origin, especially due to cerebral palsy; it is started at about 0.3 mg/kg/day in 2 to 4 doses (preferably 4), increased at weekly intervals to a usual maintenance of 0.75 to 2 mg/kg/day, without exceeding 40 mg/day before 8 years and 60 mg/day after 8 years; the tablets are not suitable below 33 kg [47]. If there is no benefit 6 to 8 weeks after reaching the maximum dose, treatment should be reassessed, and withdrawal is gradual over 1 to 2 weeks, because abrupt discontinuation can cause hallucinations, a confusional state, seizures and rebound spasticity [47]. In the United States, the oral solution Ozobax® DS is authorised only for spasticity due to multiple sclerosis and spinal cord lesions in adults (5 mg three times daily with increases every 3 days up to a maximum of 80 mg/day), with no paediatric dose; paediatric use is off-label there [48]. In Brazil there are active baclofen registrations, including Baclon® [46]. The most frequent adverse effects are sedation, drowsiness and nausea; baclofen may worsen epilepsy and should be used with caution when spasticity supports posture and gait [47, 48].

Botulinum toxin A for focal spasticity. OnabotulinumtoxinA (Botox®) blocks acetylcholine release at the neuromuscular junction. In Portugal, the paediatric indication is focal spasticity of the ankle and foot in ambulant children with cerebral palsy, from 2 years of age, as an adjunct to rehabilitation; the dose is 4 to 8 units/kg divided among the affected muscles, without exceeding 8 units/kg or 300 units per session in the lower limbs, or 10 units/kg or 340 units when both limbs are treated, within a 12-week interval, with reinjection no sooner than 12 weeks [49]. The FDA label authorises paediatric upper limb (3 to 6 units/kg, maximum 200 units) and lower limb (4 to 8 units/kg, maximum 300 units) spasticity from 2 years of age, with a total not to exceed the lesser of 10 units/kg or 340 units in 3 months [38]. The paediatric upper limb is therefore an indication in the United States and not in Portugal. Units are not interchangeable between botulinum toxin products [38]. The FDA boxed warning and the Portuguese SmPC refer to distant spread of toxin effect, with weakness, dysphagia and respiratory difficulty, probably with greater risk in children treated for spasticity, and cases of aspiration pneumonia in children with severe cerebral palsy; the most common paediatric adverse reaction in the FDA label is upper respiratory tract infection [38, 49]. In Brazil, Botox® and other botulinum toxins A have active registration [46].

Epilepsy. Seizures are infrequent and the sources describe no specific antiepileptic drugs or particular responses; treatment follows the general rules for the seizure type and epilepsy syndrome [1, 17].

No dose has been studied in NEDBA; there is no guideline with a dose for this condition, and the doses shown are those of the official labels in the general indications.

7.4. Regulatory status

The text of the Brazilian package inserts was not consulted; the table indicates only active registration in ANVISA open data.

7.5. Investigational therapies

Antisense oligonucleotides. The demonstration that p.(Arg578Cys) acts through toxic gain of function opened the way to reducing the mutant protein [25]. Zhang and colleagues tested three gapmer ASOs with phosphorothioate and 2′-O-methoxyethyl chemistry, which activate RNase H1 and do not distinguish the mutant from the normal allele. In the patient's fibroblasts and iPSC neurons, an 80–90% reduction in mutant JIP3 and a 60–70% reduction in normal JIP3 was well tolerated and restored proliferation, endosome motility and D1 receptor signalling (cyclic AMP production after agonist), reduced JNK activation and decreased neuronal death, in a dose-dependent manner; ASO2 was the most effective, with a reduction of more than 80% in mutant mRNA [25]. The authors concluded that modest allelic selectivity is sufficient, although an ASO selective for the mutant allele would be the ideal approach [25]. These results are preclinical and in vitro.

The n-Lorem Foundation, which develops personalised ASOs free of charge for patients with «nano-rare» diseases (fewer than 30 known patients worldwide), cites MAPK8IP3 as an example of toxic gain of function in which it was necessary to characterise the cellular phenotype and compare reduction of the normal and mutant alleles to define the minimum selectivity; the foundation's general rule is a selectivity of at least 5-fold, or 10-fold for ion channels [50]. Reduction of the normal allele has a theoretical limit in a disorder in which haploinsufficiency is also pathogenic [26].

Repurposed approved medicines. In a 2025 preprint, amantadine and levodopa, chosen by clinicians and by the computational repurposing tool mediKanren from FDA-approved medicines that are safe in children and available, corrected the locomotor deficits of JIP3-deficient zebrafish [51]. mediKanren had already listed MAPK8IP3-related disorder among the cases with an identified candidate therapy, without outcome data for that patient [52]. Zhang and colleagues suggest levodopa as a therapeutic possibility based on D1 receptor uncoupling [25]. None of these medicines has been tested in patients with NEDBA in a published study, there are no doses for this indication and their use outside a research protocol has no support in the sources.

7.6. Clinical trials

Secondary outcomes include seizures (EEG and diary), respiratory infections requiring treatment or admission, sleep by actigraphy, caregiver global impression of change and adverse events [19]. The registry does not state the participant's variant; the preclinical work of the same Columbia and n-Lorem group focuses on p.(Arg578Cys), but the link is not declared in the registry [19, 25]. A single-participant study without a control group, with developmental outcomes that change with age, does not allow efficacy to be estimated, and extrapolation of an ASO designed for one variant to other variants is not possible. A search of ClinicalTrials.gov for «MAPK8IP3», «JIP3» and «NEDBA» on 28 October 2026 returned only this study, with no suspended, withdrawn or negative trials. The natural history study of Columbia University and Boston Children's Hospital, which recruits through the Wolverine Foundation, is preparing outcomes for future trials [3, 4, 22].

7.7. Developments in the last 12 months

Between November 2025 and October 2026, the output on NEDBA was as follows, by level of evidence. In published clinical studies: the print version of the series of 32 individuals (February 2026) [3], the first family with dominant transmission and RNA characterisation (April 2026) [13] and an example of reclassification of an inherited variant as likely benign in a series of 822 diagnoses by paediatric clinical exome sequencing with automated interpretation [34]. In preprints not peer reviewed: the domain-specific knock-in mice with human genotype–phenotype correlation (July 2026) [26] and the preliminary association with obsessive-compulsive disorder (April 2026) [36]. In basic biology: the function of JIP3 in endolysosomal maturation in the cell body [30] and the mechanism of kinesin-1 activation by JIP3 and JIP4 [32]. In clinical research: the single-participant ASO study remains active, without results [19], and the perspective published by n-Lorem uses MAPK8IP3 as a methodological example [50]. No applications for orphan designation, opinions or authorisations were identified at the FDA or EMA; the absence of results in a search does not demonstrate the non-existence of non-public procedures.

7.8. Agents to be used with caution, pregnancy, surgery and transition

No medicine is specifically contraindicated by NEDBA in the sources consulted. Precautions derive from the comorbidities: methylphenidate and baclofen may lower the seizure threshold [44, 47], central nervous system depressants add to the sedation of baclofen [48], and botulinum toxin requires caution in children with dysphagia, significant neurological debility or a history of aspiration [49].

Pregnancy is a consideration for women with mild forms, such as the mother in the Chinese family [13]. The Portuguese SmPC does not recommend methylphenidate in pregnancy, unless delaying treatment poses a greater risk, reports a small increase in cardiac malformations with first-trimester exposure (adjusted relative risk 1.3; 95% confidence interval 1.0–1.6; about 3 additional cases per 1000 pregnancies) and describes passage into breast milk [44]. Oral baclofen taken during pregnancy can cause a neonatal withdrawal syndrome (increased tone, tremor, seizures); the FDA label recommends tapering the dose gradually and stopping before delivery, if clinically possible [48].

In the perioperative setting, methylphenidate should not be taken on the day of planned surgery, because of the risk of a sudden rise in blood pressure [44], and baclofen should not be stopped abruptly during fasting or admission [47]. The sources describe no anaesthetic risks specific to NEDBA.

Transition to adulthood is poorly documented: the oldest patients described were 29 years old [1]. In Portugal, Rubifen® has no indication in adults with ADHD, which makes it necessary to review the medicine or formulation at transition [44]. Continuity of rehabilitation, decision-making capacity and social support belong to the general planning of ID in adulthood.

8. Reproductive Options and Family Counselling

8.1. Recurrence risks by scenario

The sources do not provide a numerical estimate of the recurrence risk associated with germline mosaicism in NEDBA; it should not be presented as negligible. The same phenomenon has been demonstrated in another axonal transport gene, KIF5C, in four siblings [41].

8.2. Cascade testing and assessment of the parents

Testing the parents serves three purposes: confirming de novo occurrence, which strengthens pathogenicity [17]; identifying a heterozygous parent with a mild phenotype, such as the mother with learning difficulties in the Chinese family [13]; and, when the variant is inherited from a truly unaffected parent, weighing against the pathogenicity of a variant of uncertain significance [34]. The Chinese family shows that clinical assessment of the parents (schooling, learning, language) should accompany the molecular result [13]. When a parent is a carrier, the proband's siblings and other at-risk relatives can be tested, especially siblings with learning or language difficulties.

8.3. Prenatal diagnosis and preimplantation genetic testing

With the familial variant identified, prenatal diagnosis by analysis of chorionic villi or amniotic fluid and preimplantation genetic testing make it possible to know whether the embryo or fetus has inherited the variant; they do not allow severity to be predicted, given the variable expressivity [13]. After a child with a de novo variant, prenatal testing in a subsequent pregnancy can be discussed because of germline mosaicism [1]. No fetal ultrasound findings have been described that would allow prenatal suspicion without molecular testing.

8.4. Counselling of minors and content of the consultation

Testing of asymptomatic minor siblings is indicated when there are developmental or learning difficulties that would benefit from intervention; carrier testing for exclusively reproductive purposes can be deferred until the minor can decide. The consultation should explain dominant inheritance, the predominance of de novo variants, variable expressivity, germline mosaicism, the absence of an approved specific treatment, the ongoing personalised ASO trial and family support resources [4, 19, 21].

9. Special Considerations: Population Diversity and Epidemiology

9.1. Prevalence and what each figure measures

There is no estimate of population prevalence or incidence of NEDBA in the sources consulted, and Orphanet gives no prevalence for the non-specific syndromic ID group in which it classifies it [15]. The available figures are counts of published cases. NORD reported, in January 2025, fewer than 100 individuals described in the literature, and notes that the real number is unknown because of the novelty of the entity and unequal access to genomic testing [4]. The founding series identified 13 patients among 27,232 sequenced individuals with neurodevelopmental disorders, a proportion of about 1 in 2100 in that patient cohort, which is not a population prevalence [17]. The 2019 and 2026 series may include the same individuals, and the counts should not be added together.

9.2. Geographical distribution and ancestry

The patients in the founding series came from Brazil, Germany, Israel, Italy, the United States and the Netherlands [17]; the Iwasawa series gathered Japanese patients [24], the series of 32 was assembled in the United States [3], the case from the cerebral palsy cohort is Australian [37], the recessive form was described by a Hungarian group with North American and French collaborators [14] and the family with dominant transmission is Chinese [13]. No founder variants or frequency differences between populations have been described; the recurrent variants p.(Arg578Cys) and p.(Arg1146Cys) arose de novo in European, American and Japanese patients, which points to recurrent mutation hotspots rather than a founder effect [1, 9, 17].

Regarding Brazil, the founding series included patients referred by a Brazilian centre [17], and the published Brazilian case with the similar phenotype (NEDMAGA) is due to ZSWIM6 and not MAPK8IP3 [8]. In Portugal, the sources consulted describe no patients; the Portuguese contribution is mechanistic, with the work of i3S in Porto on motor coordination by JIP3 [31].

9.3. Special populations

Children with cerebral palsy without a sufficient perinatal cause, including those with imaging of white matter injury, are a population in which the diagnosis may be hidden [37]. Adults with ID or mild learning difficulties, especially parents of affected children, may also be undiagnosed carriers [13]. Patients with a negative exome before 2019 are another group in which reanalysis may provide the diagnosis [33]. There are no data on pregnancy in affected women or on ageing.

9.4. Reference networks, associations and information gaps

The relevant European reference network is ERN ITHACA, which covers rare malformation syndromes, ID and other neurodevelopmental disorders and maintains, through its neurodevelopmental disorders working group, the SysNDD database, in which NEDBA is registered as a definitive entity; the ERN website has no gene-specific page [16, 20]. The CureMAPK8IP3 Foundation, a non-profit organisation based in South Carolina (USA), maintains a family registry, refers families to the natural history study, organises a biennial family conference (the 2026 one, hybrid, in Boston, on 6 and 7 November) and supports research with patient iPSCs at the University of Alabama at Birmingham [21]. The Wolverine Foundation, of Greenwich (Connecticut), funds research into mechanisms and therapies and refers families to the natural history study [4, 22]. NORD publishes a disease report aimed at families and professionals [4]. The sources consulted identify no specific associations in Portugal or Brazil.

Three primary articles were accessible only as abstracts: the series of 32 individuals (frequencies by sign, longitudinal adaptive functioning and management), the Japanese series (individual data, known through OMIM) and the recessive case (variants and course) [3, 14, 24]. The data presented from these studies are limited to what is given in the abstracts and the OMIM entry.

10. Illustrative Clinical Case

10.1. Recurrent variant p.(Arg578Cys) with brain abnormalities (Platzer 2019, individual 8)

A girl assessed at 5 years of age in the founding series had psychomotor developmental delay from early childhood and severe ID [17]. On examination she had muscular hypotonia associated with spasticity, microcephaly of −2.5 standard deviations and short stature, and had a single generalised epileptic seizure. Brain MRI at 3 years (midsagittal and axial T2 sequences) showed a very thin corpus callosum with hypoplasia of the rostrum and splenium, cerebral atrophy with widened sulci, reduced white matter volume and cerebellar atrophy (Figure 1, images 8.1 and 8.2, of Platzer et al. 2019: https://pmc.ncbi.nlm.nih.gov/articles/PMC6369540/figure/fig1/). Exome sequencing identified the heterozygous de novo variant NM_015133.4:c.1732C>T p.(Arg578Cys), which corresponds in MANE to NM_001318852.2:c.1735C>T p.(Arg579Cys), absent from gnomAD [12, 17].

A second girl in the same series, unrelated and with the same variant, had at 6 years moderate ID, hypotonia and spasticity, microcephaly of −2.5 standard deviations, short stature and one generalised seizure, and MRI at 5 years showed a thin corpus callosum, cerebral and cerebellar atrophy and white matter loss (images 9.1 and 9.2 of the same figure) [17]. The similarity between the two, and with the three Japanese patients carrying the same variant, supported the first genotype–phenotype correlation of the disorder [1, 17]. The same variant was later studied in cells from a patient with dystonia and motor delay, and showed a toxic gain-of-function mechanism; it is also the variant of the preclinical work that preceded the personalised ASO study [19, 25]. The report describes no treatment.

The case shows the elements that point towards NEDBA in practice: a child with severe global delay, a combination of hypotonia and spasticity, microcephaly and an MRI with a thin corpus callosum and atrophy, in whom the diagnosis depends on exome sequencing and interpretation requires conversion of the historical nomenclature to MANE.

10.2. Inherited splicing variant with variable expressivity (Yi 2026)

A Chinese boy, the second child of non-consanguineous parents, was born at term after an uneventful pregnancy [13]. He achieved head control at 5 months, sat at 8 and walked at 21 months. At 2 years he did not respond to his name, avoided eye contact and did not follow commands, and received a diagnosis of autism, starting rehabilitation with limited improvement. On the Gesell scale, at 3 years and 11 months, the developmental quotients were 51 in the gross motor domain, 26 in fine motor, 25 in adaptive and 33 in language and the social domain. He was 98 cm tall (−1.2 SD), weighed 17 kg and had a head circumference of 52.5 cm (+1.7 SD), used short sentences without spontaneous language and had an inguinal hernia, without malformations or dysmorphism. Chest and spine radiographs, echocardiogram, abdominal ultrasound, EEG and brain MRI were normal, and none of the affected relatives had hypotonia, spasticity, ataxia or seizures.

Exome sequencing identified the variant NM_001318852.2:c.2630-2_2630-1del, at the intron 21 acceptor, inherited from the mother and, through her, from the maternal grandmother [13]. The mother completed the 9th year of schooling with mild learning difficulties and was never employed; the sister had language delay and learning difficulties; the grandmother, a carrier, had not walked since childhood, which the family attributed to rheumatoid arthritis, and her developmental history was unknown. Leucocyte RNA sequencing showed aberrant splicing in 44 to 58% of the reads of the three carriers studied, through activation of a cryptic acceptor in exon 22 with a 44-nucleotide deletion, and the variant was classified as likely pathogenic [13]. Two other variants found, in NFIB (from the healthy father) and in EP300 (from the healthy grandfather), were classified as of uncertain significance. Figure 1 of the article shows the three-generation pedigree with the phenotype and genotype of each member (A), the Sanger electropherograms with the deletion marked (B) and the spectrum of published MAPK8IP3 variants on the protein, with the RH1, RH2, leucine zipper and WD40 domains and the new variant highlighted (C); Figure 2 shows the sashimi plot of RNA sequencing with loss of part of exon 22 in the carriers (A), the 409 bp mutant RT-PCR product present in the relatives and absent in controls on 2.5% agarose gel (B), the cDNA chromatogram with double peaks from r.2630 onwards (C) and TA clone sequencing with the aberrant transcript lacking 44 bp of exon 22, found in 8 of 18 clones (D) (https://pmc.ncbi.nlm.nih.gov/articles/PMC13101558/) [13]. Only the proband had a standardised developmental assessment and MRI; the mother, the sister and the grandmother did not undergo formal cognitive or neurological assessment or MRI, and the mother reported, without confirmation, macrocephaly, speech disorder and oesophageal malformations in maternal relatives who were not examined, so the intrafamilial variability should be interpreted with caution [13].

The case contradicts the idea that NEDBA is always de novo, shows that a loss-of-function variant can cause anything from mild learning difficulties to severe global delay with ASD in the same family and illustrates the value of RNA studies in classifying splicing variants [13].

11. Prognosis and Long-Term Follow-up

11.1. Prognosis

The prognostic data come from cross-sectional series and from adaptive follow-up not yet accessible in full text. The disorder is, as a rule, a non-progressive developmental encephalopathy: none of the 13 patients in the founding series regressed [17] and the knock-in mice showed no degeneration up to 12 months [26]; one case with regression at 30 months is the exception described [33]. In the series of 32 individuals, almost all acquired independent walking, but with reduced speed and a broad base, and general conceptual ability had a mean of 62.5 ± 26.5 [3]. Carriers of the recurrent missense variants have a worse functional prognosis: in the Japanese series, the three oldest patients (up to 29 years) never walked and were wheelchair-dependent [1], and in a 2026 review ID was severe or profound in 70% of carriers of p.(Arg578Cys) or p.(Arg1146Cys), with language limited to single words or absent [13]. Loss-of-function variants are associated with milder pictures, ranging from learning difficulties in a functional adult to severe global delay [3, 13]. There are no published data on life expectancy or on the course in later adulthood; the oldest patients described were 29 years old [1].

11.2. Surveillance

There are no published surveillance recommendations for NEDBA. The table gathers the domains supported by the sources; the clinical intervals follow general practice in the follow-up of ID and its comorbidities and do not derive from a specific guideline, with the exception of those for methylphenidate and botulinum toxin, which appear in the labels.

12. Summary and Final Considerations

NEDBA is an autosomal dominant neurodevelopmental disorder caused by heterozygous MAPK8IP3 variants, almost always de novo, described in 2019 and with definitive gene–disease validity. The gene encodes JIP3, a scaffold protein of the JNK pathway and adaptor of the axonal transport motors that coordinates kinesin-1 and dynein and regulates the trafficking and maturation of lysosomes and endosomes. The acronym NEDMAGA does not belong to this disorder: it designates the disorder caused by ZSWIM6, the closest clinical differential.

The clinical picture combines global delay from infancy, mild to profound ID, poor or absent speech, hypotonia, spasticity or ataxia, strabismus, microcephaly, attention deficit and ASD. About half to two-thirds of patients have MRI abnormalities, of which a thin corpus callosum is the most frequent (62.1% in the series of 32 individuals), followed by cerebral and cerebellar atrophy and perisylvian polymicrogyria; a normal MRI is compatible with the diagnosis. Seizures are infrequent, regression is exceptional and there is no characteristic facial phenotype. The recurrent missense variants p.(Arg578Cys) and p.(Arg1146Cys), in MANE p.(Arg579Cys) and p.(Arg1147Cys), are associated with the most severe pictures, and the former acts through toxic gain of function; loss-of-function variants tend to cause milder pictures and may be inherited from parents with learning difficulties.

Diagnosis is by exome or genome sequencing, with careful conversion of the historical nomenclature, testing of the parents with assessment of their phenotype and RNA studies for splicing variants; a negative exome, especially one before 2019, does not exclude the disorder. Treatment is supportive and symptomatic, based on rehabilitation, and medicines for ADHD and spasticity are used within the general indications, with label doses. The personalised ASO under study in a single participant, following the in vitro demonstration of reversal of the toxic gain of function, is the first targeted therapy in clinical research, without published results. Counselling should address the 50% risk for the offspring of carriers, variable expressivity and germline mosaicism after a de novo variant. The main gaps are the absence of prevalence data, of data in adulthood and of validated surveillance recommendations.

13. References

1. Online Mendelian Inheritance in Man. #618443 Neurodevelopmental disorder with or without variable brain abnormalities; NEDBA [Internet]. Baltimore: Johns Hopkins University; created 2019-05-22, edited 2021-12-27 [cited 2026-10-28]. Available from: https://omim.org/entry/618443
2. National Center for Biotechnology Information. MedGen: Neurodevelopmental disorder with or without variable brain abnormalities; NEDBA (UID 1675664; C5193102) [Internet]. Bethesda: NLM [cited 2026-10-28]. Available from: https://www.ncbi.nlm.nih.gov/medgen/1675664
3. Sudnawa KK, Geltzeiler A, Kanner CH, Zreibe K, Pini N, Tam C, et al. Comprehensive clinical characteristics, longitudinal adaptive functioning, and electroencephalogram activity in MAPK8IP3-related neurodevelopmental disorder. Clin Genet. 2026;109(2):233-42. doi:10.1111/cge.70032. PMID 40734308.
4. National Organization for Rare Disorders. MAPK8IP3-Related Neurodevelopmental Disorder [Internet]. Danbury (CT): NORD; 2021, updated 2025-01-24 [cited 2026-10-28]. Available from: https://rarediseases.org/rare-diseases/mapk8ip3-related-neurodevelopmental-disorder/
5. National Center for Biotechnology Information. ClinVar: MAPK8IP3[gene] [Internet; cited 2026-10-28]. Available from: https://www.ncbi.nlm.nih.gov/clinvar/?term=MAPK8IP3%5Bgene%5D
6. National Center for Biotechnology Information. MedGen: Neurodevelopmental disorder with movement abnormalities, abnormal gait, and autistic features (UID 1647077; C4693405; OMIM 617865) [Internet]. Bethesda: NLM [cited 2026-10-28]. Available from: https://www.ncbi.nlm.nih.gov/medgen/1647077
7. Palmer EE, Kumar R, Gordon CT, Shaw M, Hubert L, Carroll R, et al. A recurrent de novo nonsense variant in ZSWIM6 results in severe intellectual disability without frontonasal or limb malformations. Am J Hum Genet. 2017;101(6):995-1005. doi:10.1016/j.ajhg.2017.10.009. PMID 29198722; PMCID PMC5812890.
8. Corazza LA, Dousseau GC, de Freitas JL, Torres IA, Rocha MSG. A case of NEDMAGA: neurodevelopmental disorder with movement abnormalities, abnormal gait, and autistic features. Mov Disord Clin Pract. 2024;11(2):181-3. doi:10.1002/mdc3.13954. PMID 38386481; PMCID PMC10883404.
9. Online Mendelian Inheritance in Man. Entry 605431 (gene): Mitogen-activated protein kinase 8-interacting protein 3; MAPK8IP3 [Internet]. Baltimore: Johns Hopkins University [cited 2026-10-28]. Available from: https://omim.org/entry/605431
10. GeneGuidance. MAPK8IP3 — mitogen-activated protein kinase 8-interacting protein 3 [database]. Curation of 31 Aug 2026; version of 26 Sep 2026 (accessed 28 Oct 2026).
11. ClinGen. MAPK8IP3 (HGNC:6884) — gene-disease validity: NEDBA (MONDO:0032755), AD, Definitive, Syndromic Disorders GCEP, 2022-05-22 [Internet; cited 2026-10-28]. Available from: https://search.clinicalgenome.org/kb/genes/HGNC:6884
12. Freeman PJ, Hart RK, Gretton LJ, Brookes AJ, Dalgleish R. VariantValidator: accurate validation, mapping, and formatting of sequence variation descriptions. Hum Mutat. 2018;39(1):61-8. doi:10.1002/humu.23348. PMID 28967166. Tool: https://variantvalidator.org (GRCh38, accessed 2026-10-28).
13. Yi S, Yang Q, Song X, Huang L, Yi S, Qin Z, et al. Characterization of a novel inherited splice-site variant in MAPK8IP3 expands the genetic and phenotypic spectrum of neurodevelopmental disorders. Biochem Biophys Rep. 2026;46:102582. doi:10.1016/j.bbrep.2026.102582. PMID 42027429; PMCID PMC13101558.
14. Kárteszi J, Ziegler A, Tihanyi M, Elmont B, Zhang Y, Patócs B, et al. Compound heterozygous variants in MAPK8IP3 were detected in severe congenital hypotonia mimicking lethal spinal muscular atrophy. Am J Med Genet A. 2023;191(9):2428-32. doi:10.1002/ajmg.a.63340. PMID 37462082.
15. Orphanet. Non-specific syndromic intellectual disability (ORPHA:528084) [Internet]. Paris: INSERM; 2026 [updated 2026-07-02; cited 2026-10-28]. Available from: https://www.orpha.net/en/disease/detail/528084
16. SysNDD. MAPK8IP3 entities 2743 and 4091 [Internet database; cited 2026-10-28]. Available from: https://sysndd.dbmr.unibe.ch/
17. Platzer K, Sticht H, Edwards SL, Allen W, Angione KM, Bonati MT, et al. De novo variants in MAPK8IP3 cause intellectual disability with variable brain anomalies. Am J Hum Genet. 2019;104(2):203-12. doi:10.1016/j.ajhg.2018.12.008. PMID 30612693; PMCID PMC6369540.
18. Manickam K, McClain MR, Demmer LA, Biswas S, Kearney HM, Malinowski J, et al. Exome and genome sequencing for pediatric patients with congenital anomalies or intellectual disability: an evidence-based clinical guideline of the American College of Medical Genetics and Genomics (ACMG). Genet Med. 2021;23(11):2029-37. doi:10.1038/s41436-021-01242-6. PMID 34211152.
19. n-Lorem Foundation. An open-label single center, single participant study of an experimental antisense oligonucleotide treatment for neurodevelopmental disorder with or without variable brain abnormalities (NEDBA) due to MAPK8IP3 mutation. ClinicalTrials.gov identifier NCT07197294 [Internet; updated 2025-09-29; cited 2026-10-28]. Available from: https://clinicaltrials.gov/study/NCT07197294
20. ERN ITHACA — European Reference Network on Rare Malformation Syndromes, Intellectual and Other Neurodevelopmental Disorders [Internet; cited 2026-10-28]. Available from: https://ern-ithaca.eu/
21. CureMAPK8IP3 Foundation [Internet; cited 2026-10-28]. Available from: https://curemapk8ip3.org/
22. The Wolverine Foundation [Internet; cited 2026-10-28]. Available from: https://www.wolvfdn.com
23. Kelkar N, Delmotte MH, Weston CR, Barrett T, Sheppard BJ, Flavell RA, et al. Morphogenesis of the telencephalic commissure requires scaffold protein JNK-interacting protein 3 (JIP3). Proc Natl Acad Sci U S A. 2003;100(17):9843-8. doi:10.1073/pnas.1733944100. PMID 12897243.
24. Iwasawa S, Yanagi K, Kikuchi A, Kobayashi Y, Haginoya K, Matsumoto H, et al. Recurrent de novo MAPK8IP3 variants cause neurological phenotypes. Ann Neurol. 2019;85(6):927-33. doi:10.1002/ana.25481. PMID 30945334.
25. Zhang W, Mittal S, Thomas R, Foroughishafiei A, Nunes Bastos R, Chung WK, et al. A toxic gain-of-function variant in MAPK8IP3 provides insights into JIP3 cellular roles. JCI Insight. 2025;10(8):e187199. doi:10.1172/jci.insight.187199. PMID 40111412; PMCID PMC12016931.
26. Crowder CM, Watkins LR, Geltzeiler A, et al. Pathogenic MAPK8IP3 variants drive distinct motor and behavioral phenotypes in humans and mice [preprint]. bioRxiv. 2026 Jul 24. doi:10.64898/2026.07.20.739682. Preprint not peer reviewed.
27. Saudubray JM, Baumgartner MR, García-Cazorla Á, Walter J, editors. Inborn Metabolic Diseases: Diagnosis and Treatment. 7th ed. Berlin: Springer; 2022. Chapter 44, Disorders of Cellular Trafficking, section 44.2.2.
28. Gowrishankar S, Wu Y, Ferguson SM. Impaired JIP3-dependent axonal lysosome transport promotes amyloid plaque pathology. J Cell Biol. 2017;216(10):3291-305. doi:10.1083/jcb.201612148. PMID 28784610.
29. Gowrishankar S, Lyons L, Rafiq NM, Roczniak-Ferguson A, De Camilli P, Ferguson SM. Overlapping roles of JIP3 and JIP4 in promoting axonal transport of lysosomes in human iPSC-derived neurons. Mol Biol Cell. 2021;32(11):1094-103. doi:10.1091/mbc.e20-06-0382. PMID 33788575.
30. Krout M, Gowrishankar S. Ultrastructural studies reveal a new role for JIP3/MAPK8IP3 in endo-lysosomal maturation within the neuronal cell body. Mol Biol Cell. 2026;37(6):ar63. doi:10.1091/mbc.e25-05-0248-t. PMID 42126962.
31. Celestino R, Gama JB, Castro-Rodrigues AF, Barbosa DJ, Rocha H, d'Amico EA, et al. JIP3 interacts with dynein and kinesin-1 to regulate bidirectional organelle transport. J Cell Biol. 2022;221(8):e202110057. doi:10.1083/jcb.202110057. PMID 35829703.
32. Vilela F, Chenon M, Campanacci V, Alik A, Velours C, Andreani J, et al. Insights into how JIP3 and JIP4 adaptor proteins relieve the auto-inhibited form of the kinesin-1 motor. J Biol Chem. 2026;302(9):113355. doi:10.1016/j.jbc.2026.113355. PMID 42486258.
33. Nomakuchi TT, Teferedegn EY, Li D, Muirhead KJ, Dubbs H, Leonard J, et al. Utility of genome sequencing in exome-negative pediatric patients with neurodevelopmental phenotypes. Am J Med Genet A. 2024;194(12):e63817. doi:10.1002/ajmg.a.63817. PMID 39031459; PMCID PMC11540733.
34. Pan Y, Danley P, Kramer T, Noruzinia M, Buser K, Yatsenko AN, et al. Artificial intelligence-assisted clinical exome sequencing: insights and outcomes from 822 pediatric diagnoses. Genet Med Open. 2026;4:104488. doi:10.1016/j.gimo.2026.104488. PMID 42701435; PMCID PMC13545448.
35. Silvaieh S, König T, Wurm R, Parvizi T, Berger-Sieczkowski E, Goeschl S, et al. Comprehensive genetic screening of early-onset dementia patients in an Austrian cohort—suggesting new disease-contributing genes. Hum Genomics. 2023;17(1):55. doi:10.1186/s40246-023-00499-z. PMID 37330543; PMCID PMC10276391.
36. Ormond C, Cap M, Chang YC, Ryan N, Chavira D, Williams K, et al. Rare protein-disrupting variants in NPY5R, DLGAP1 and MAPK8IP3 segregate with OCD in two multiplex pedigrees [preprint]. medRxiv. 2026 Apr 22. doi:10.64898/2026.04.21.26350600. PMID 42078338.
37. Berry JG, Taranath A, Goetti R, Farrar MA, Fiori S, Pham HD, et al. Genetic diagnostic yield by MRI pattern in children with cerebral palsy: a population-based study. EBioMedicine. 2025;122:106013. doi:10.1016/j.ebiom.2025.106013. PMID 41202470; PMCID PMC12790587.
38. U.S. Food and Drug Administration. BOTOX (onabotulinumtoxinA) for injection — prescribing information [Internet]. BLA103000; AbbVie/Allergan; label effective date 2 Feb 2026. DailyMed setid a6f42753-d31e-4b23-8da0-4b990fee4b38. Available from: https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=a6f42753-d31e-4b23-8da0-4b990fee4b38 (accessed 28 Oct 2026).
39. Richards S, Aziz N, Bale S, Bick D, Das S, Gastier-Foster J, et al. Standards and guidelines for the interpretation of sequence variants: a joint consensus recommendation of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology. Genet Med. 2015;17(5):405-24. doi:10.1038/gim.2015.30. PMID 25741868.
40. GeneGuidance. KIF1A [database]. Version of 26 Sep 2026 (accessed 28 Oct 2026).
41. GeneGuidance. KIF5C [database]. Version of 26 Sep 2026 (accessed 28 Oct 2026).
42. GeneGuidance. DYNC1H1 [database]. Version of 26 Sep 2026 (accessed 28 Oct 2026).
43. GeneGuidance. KIF5A [database]. Version of 26 Sep 2026 (accessed 28 Oct 2026).
44. INFARMED. Rubifen 5 mg, 10 mg e 20 mg comprimidos (metilfenidato) — Resumo das Características do Medicamento [Internet]. Marketing authorisation holder: Laboratorios Rubió; text revised September 2025. Infomed med_id 36077. Available from: https://extranet.infarmed.pt/INFOMED-fo/ (accessed 28 Oct 2026).
45. U.S. Food and Drug Administration. Ritalin (methylphenidate hydrochloride) tablets — prescribing information [Internet]. NDA010187; label effective date 5 Feb 2025 (adverse reaction reporting: Sandoz Inc.). DailyMed setid d6fb2750-cdab-4749-ba0d-7534840a5892. Available from: https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=d6fb2750-cdab-4749-ba0d-7534840a5892 (accessed 28 Oct 2026).
46. Agência Nacional de Vigilância Sanitária (ANVISA). Open data: registered medicines (DADOS_ABERTOS_MEDICAMENTOS.csv) — methylphenidate, baclofen and botulinum toxin A [Internet]. Brasília: ANVISA. Available from: https://dados.anvisa.gov.br/dados/DADOS_ABERTOS_MEDICAMENTOS.csv (accessed 28 Oct 2026).
47. INFARMED. Lioresal 10 mg e 25 mg comprimidos (baclofeno) — Resumo das Características do Medicamento [Internet]. Marketing authorisation holder: Novartis Farma. Infomed med_id 5074. Available from: https://extranet.infarmed.pt/INFOMED-fo/ (accessed 28 Oct 2026).
48. U.S. Food and Drug Administration. Ozobax DS (baclofen) oral solution — prescribing information [Internet]. NDA208193; Metacel Pharmaceuticals; label effective date 13 Mar 2026. DailyMed setid d8ddcec5-debe-42b4-a803-e65917cc6a54. Available from: https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=d8ddcec5-debe-42b4-a803-e65917cc6a54 (accessed 28 Oct 2026).
49. INFARMED. Botox 100 Unidades pó para solução injetável (toxina botulínica A) — Resumo das Características do Medicamento [Internet]. Marketing authorisation holder: AbbVie, Lda. Infomed med_id 29406. Available from: https://extranet.infarmed.pt/INFOMED-fo/download-ficheiro.xhtml?med_id=29406&tipo_doc=rcm (accessed 28 Oct 2026).
50. Crooke ST, Glass S, Gleeson JG, Mignon L, Skourti-Stathaki K, Douville J, et al. Addressing the needs of nano-rare patients: the n-Lorem experience. Nucleic Acids Res. 2026;54(10):gkag504. doi:10.1093/nar/gkag504. PMID 42227334; PMCID PMC13227102.
51. Foksinska A, Souder P, Smith G, Travis K, et al. Anti-Parkinsonian drugs rescue locomotor deficits in JIP3 knockout zebrafish: implications for treating patients with MAPK8IP3-related neurodevelopmental disorders [preprint]. bioRxiv. 2025 Feb 6. doi:10.1101/2025.02.02.636112. Preprint not peer reviewed.
52. Foksinska A, Crowder CM, Crouse AB, Henrikson J, Byrd WE, Rosenblatt G, et al. The precision medicine process for treating rare disease using the artificial intelligence tool mediKanren. Front Artif Intell. 2022;5:910216. doi:10.3389/frai.2022.910216. PMID 36248623; PMCID PMC9562701.